mewin/glm
1
0
Fork 0
Code Releases Activity

Fixed merge

Browse Source
This commit is contained in:
Christophe Riccio
2013-01-31 15:01:48 +01:00
parent 739ab3529c 7848efdba9
commit b5607d0018
54 changed files with 1999 additions and 1313 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
glm/core/_detail.hpp
View File

@@ -445,6 +445,13 @@ namespace detail
# define GLM_RESTRICT __declspec(restrict)
# define GLM_RESTRICT_VAR __restrict
# define GLM_CONSTEXPR
#elif(GLM_COMPILER & GLM_COMPILER_INTEL)
# define GLM_DEPRECATED
# define GLM_ALIGN(x) __declspec(align(x))
# define GLM_ALIGNED_STRUCT(x) __declspec(align(x)) struct
# define GLM_RESTRICT
# define GLM_RESTRICT_VAR __restrict
# define GLM_CONSTEXPR
#elif((GLM_COMPILER & (GLM_COMPILER_GCC | GLM_COMPILER_LLVM_GCC)) && (GLM_COMPILER >= GLM_COMPILER_GCC31))
# define GLM_DEPRECATED __attribute__((__deprecated__))
# define GLM_ALIGN(x) __attribute__((aligned(x)))

4
glm/core/func_common.hpp
View File

@@ -278,8 +278,8 @@ namespace glm
/// you would want a threshold function with a smooth
/// transition. This is equivalent to:
/// genType t;
/// t = clamp ((x <EFBFBD> edge0) / (edge1 <EFBFBD> edge0), 0, 1);
/// return t * t * (3 <EFBFBD> 2 * t);
/// t = clamp ((x - edge0) / (edge1 - edge0), 0, 1);
/// return t * t * (3 - 2 * t);
/// Results are undefined if edge0 >= edge1.
///
/// @tparam genType Floating-point scalar or vector types.

11
glm/core/func_common.inl
View File

@@ -275,7 +275,7 @@ namespace detail
//// Only valid if (INT_MIN <= x-y <= INT_MAX)
//// min(x,y)
//r = y + ((x - y) & ((x - y) >> (sizeof(int) *
//CHAR_BIT <EFBFBD> 1)));
//CHAR_BIT - 1)));
//// max(x,y)
//r = x - ((x - y) & ((x - y) >> (sizeof(int) *
//CHAR_BIT - 1)));
@@ -615,7 +615,7 @@ namespace detail
genType const & x
)
{
GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'mix' only accept floating-point inputs");
GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'step' only accept floating-point inputs");
return x < edge ? genType(0) : genType(1);
}
@@ -707,7 +707,7 @@ namespace detail
genType const & x
)
{
GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'mix' only accept floating-point inputs");
GLM_STATIC_ASSERT(detail::type<genType>::is_float, "'smoothstep' only accept floating-point inputs");
genType tmp = clamp((x - edge0) / (edge1 - edge0), genType(0), genType(1));
return tmp * tmp * (genType(3) - genType(2) * tmp);
@@ -811,6 +811,8 @@ namespace detail
# else
return std::isnan(x);
# endif
# elif(GLM_COMPILER & GLM_COMPILER_CUDA)
return isnan(x) != 0;
# else
return std::isnan(x);
# endif
@@ -866,6 +868,9 @@ namespace detail
# else
return std::isinf(x);
# endif
# elif(GLM_COMPILER & GLM_COMPILER_CUDA)
// http://developer.download.nvidia.com/compute/cuda/4_2/rel/toolkit/docs/online/group__CUDA__MATH__DOUBLE_g13431dd2b40b51f9139cbb7f50c18fab.html#g13431dd2b40b51f9139cbb7f50c18fab
return isinf(double(x)) != 0;
# else
return std::isinf(x);
# endif

4
glm/core/func_integer.hpp
View File

@@ -105,7 +105,7 @@ namespace glm
/// returning them in the least significant bits of the result.
/// For unsigned data types, the most significant bits of the
/// result will be set to zero. For signed data types, the
/// most significant bits will be set to the value of bit offset + base <EFBFBD> 1.
/// most significant bits will be set to the value of bit offset + base - 1.
///
/// If bits is zero, the result will be zero. The result will be
/// undefined if offset or bits is negative, or if the sum of
@@ -125,7 +125,7 @@ namespace glm
/// Returns the insertion the bits least-significant bits of insert into base.
///
/// The result will have bits [offset, offset + bits - 1] taken
/// from bits [0, bits <EFBFBD> 1] of insert, and all other bits taken
/// from bits [0, bits - 1] of insert, and all other bits taken
/// directly from the corresponding bits of base. If bits is
/// zero, the result will simply be base. The result will be
/// undefined if offset or bits is negative, or if the sum of

34
glm/core/func_packing.inl
View File

@@ -136,12 +136,42 @@ namespace glm
GLM_FUNC_QUALIFIER double packDouble2x32(detail::tvec2<detail::uint32> const & v)
{
return *(double*)&v;
struct uint32_pair
{
detail::uint32 x;
detail::uint32 y;
};
union helper
{
uint32_pair input;
double output;
} Helper;
Helper.input.x = v.x;
Helper.input.y = v.y;
return Helper.output;
//return *(double*)&v;
}
GLM_FUNC_QUALIFIER detail::tvec2<uint> unpackDouble2x32(double const & v)
{
return *(detail::tvec2<uint>*)&v;
struct uint32_pair
{
detail::uint32 x;
detail::uint32 y;
};
union helper
{
double input;
uint32_pair output;
} Helper;
Helper.input = v;
return detail::tvec2<uint>(Helper.output.x, Helper.output.y);
}
GLM_FUNC_QUALIFIER uint packHalf2x16(detail::tvec2<float> const & v)

2
glm/core/intrinsic_common.inl
View File

@@ -62,7 +62,7 @@ namespace detail{
static const ieee754_QNAN absMask;
static const __m128 GLM_VAR_USED abs4Mask = _mm_set_ps1(absMask.f);
static const __m128 GLM_VAR_USED _epi32_sign_mask = _mm_castsi128_ps(_mm_set1_epi32(0x80000000));
static const __m128 GLM_VAR_USED _epi32_sign_mask = _mm_castsi128_ps(_mm_set1_epi32(static_cast<int>(0x80000000)));
//static const __m128 GLM_VAR_USED _epi32_inv_sign_mask = _mm_castsi128_ps(_mm_set1_epi32(0x7FFFFFFF));
//static const __m128 GLM_VAR_USED _epi32_mant_mask = _mm_castsi128_ps(_mm_set1_epi32(0x7F800000));
//static const __m128 GLM_VAR_USED _epi32_inv_mant_mask = _mm_castsi128_ps(_mm_set1_epi32(0x807FFFFF));

8
glm/core/intrinsic_matrix.inl
View File

@@ -46,7 +46,7 @@ GLM_FUNC_QUALIFIER matType sse_comp_mul_ps
out[3] = _mm_mul_ps(in1[3], in2[3]);
}
GLM_FUNC_QUALIFIER void sse_add_ps(__m128 in1[4], __m128 in2[4], __m128 out[4])
GLM_FUNC_QUALIFIER void sse_add_ps(__m128 const in1[4], __m128 const in2[4], __m128 out[4])
{
{
out[0] = _mm_add_ps(in1[0], in2[0]);
@@ -56,7 +56,7 @@ GLM_FUNC_QUALIFIER void sse_add_ps(__m128 in1[4], __m128 in2[4], __m128 out[4])
}
}
GLM_FUNC_QUALIFIER void sse_sub_ps(__m128 in1[4], __m128 in2[4], __m128 out[4])
GLM_FUNC_QUALIFIER void sse_sub_ps(__m128 const in1[4], __m128 const in2[4], __m128 out[4])
{
{
out[0] = _mm_sub_ps(in1[0], in2[0]);
@@ -66,7 +66,7 @@ GLM_FUNC_QUALIFIER void sse_sub_ps(__m128 in1[4], __m128 in2[4], __m128 out[4])
}
}
GLM_FUNC_QUALIFIER __m128 sse_mul_ps(__m128 m[4], __m128 v)
GLM_FUNC_QUALIFIER __m128 sse_mul_ps(__m128 const m[4], __m128 v)
{
__m128 v0 = _mm_shuffle_ps(v, v, _MM_SHUFFLE(0, 0, 0, 0));
__m128 v1 = _mm_shuffle_ps(v, v, _MM_SHUFFLE(1, 1, 1, 1));
@@ -85,7 +85,7 @@ GLM_FUNC_QUALIFIER __m128 sse_mul_ps(__m128 m[4], __m128 v)
return a2;
}
GLM_FUNC_QUALIFIER __m128 sse_mul_ps(__m128 v, __m128 m[4])
GLM_FUNC_QUALIFIER __m128 sse_mul_ps(__m128 v, __m128 const m[4])
{
__m128 i0 = m[0];
__m128 i1 = m[1];

9
glm/core/setup.hpp
View File

@@ -36,7 +36,7 @@
#define GLM_VERSION_MAJOR 0
#define GLM_VERSION_MINOR 9
#define GLM_VERSION_PATCH 4
#define GLM_VERSION_REVISION 0
#define GLM_VERSION_REVISION 2
///////////////////////////////////////////////////////////////////////////////////////////////////
// Platform
@@ -536,6 +536,13 @@
# define GLM_ARCH GLM_ARCH_PURE
#endif
// With MinGW-W64, including intrinsic headers before intrin.h will produce some errors. The problem is
// that windows.h (and maybe other headers) will silently include intrin.h, which of course causes problems.
// To fix, we just explicitly include intrin.h here.
#if defined(__MINGW32__) && (GLM_ARCH != GLM_ARCH_PURE)
# include <intrin.h>
#endif
//#if(GLM_ARCH != GLM_ARCH_PURE)
#if(GLM_ARCH & GLM_ARCH_AVX2)
# include <immintrin.h>

2
glm/core/type_half.inl
View File

@@ -155,7 +155,7 @@ namespace detail
// We convert f to a half zero.
//
return 0;
return hdata(s);
}
//

20
glm/gtc/epsilon.hpp
View File

@@ -57,11 +57,11 @@ namespace glm
/// Returns the component-wise compare of |x - y| < epsilon.
/// @see gtc_epsilon
template <typename genType>
typename genType::boolType epsilonEqual(
genType const & x,
genType const & y,
typename genType::value_type const & epsilon);
typename genType::boolType epsilonEqual(
genType const & x,
genType const & y,
typename genType::value_type const & epsilon);
/// Returns the component-wise compare of |x - y| < epsilon.
/// @see gtc_epsilon
template <typename genType>
@@ -73,11 +73,11 @@ namespace glm
/// Returns the component-wise compare of |x - y| < epsilon.
/// @see gtc_epsilon
template <typename genType>
typename genType::boolType epsilonNotEqual(
genType const & x,
genType const & y,
typename genType::value_type const & epsilon);
typename genType::boolType epsilonNotEqual(
genType const & x,
genType const & y,
typename genType::value_type const & epsilon);
/// Returns the component-wise compare of |x - y| >= epsilon.
/// @see gtc_epsilon
template <typename genType>

30
glm/gtc/epsilon.inl
View File

@@ -269,21 +269,6 @@ namespace glm
abs(x.w - y.w) < epsilon);
}
template <typename valType>
GLM_FUNC_QUALIFIER detail::tvec4<bool> epsilonEqual
(
detail::tquat<valType> const & x,
detail::tquat<valType> const & y,
detail::tquat<valType> const & epsilon
)
{
return detail::tvec4<bool>(
abs(x.x - y.x) < epsilon.x,
abs(x.y - y.y) < epsilon.y,
abs(x.z - y.z) < epsilon.z,
abs(x.w - y.w) < epsilon.w);
}
template <typename valType>
GLM_FUNC_QUALIFIER detail::tvec4<bool> epsilonNotEqual
(
@@ -298,19 +283,4 @@ namespace glm
abs(x.z - y.z) >= epsilon,
abs(x.w - y.w) >= epsilon);
}
template <typename valType>
GLM_FUNC_QUALIFIER detail::tvec4<bool> epsilonNotEqual
(
detail::tquat<valType> const & x,
detail::tquat<valType> const & y,
detail::tquat<valType> const & epsilon
)
{
return detail::tvec4<bool>(
abs(x.x - y.x) >= epsilon.x,
abs(x.y - y.y) >= epsilon.y,
abs(x.z - y.z) >= epsilon.z,
abs(x.w - y.w) >= epsilon.w);
}
}//namespace glm

10
glm/gtc/matrix_transform.hpp
View File

@@ -87,8 +87,8 @@ namespace glm
///
/// @param m Input matrix multiplied by this rotation matrix.
/// @param angle Rotation angle expressed in radians if GLM_FORCE_RADIANS is define or degrees otherwise.
/// @param axis Rotation axis.
/// @tparam T Value type used to build the matrix. Currently supported: half (not recommanded), float or double.
/// @param axis Rotation axis, recommanded to be normalized.
/// @tparam T Value type used to build the matrix. Supported: half, float or double.
/// @see gtc_matrix_transform
/// @see gtx_transform
/// @see - rotate(T angle, T x, T y, T z)
@@ -143,7 +143,7 @@ namespace glm
/// @param top
/// @tparam T Value type used to build the matrix. Currently supported: half (not recommanded), float or double.
/// @see gtc_matrix_transform
/// @see - glm::ortho(T const & left, T const & right, T const & bottom, T const & top, T const & zNear, T const & zFar)
/// @see - glm::ortho(T const & left, T const & right, T const & bottom, T const & top, T const & zNear, T const & zFar)
template <typename T>
detail::tmat4x4<T> ortho(
T const & left,
@@ -209,7 +209,7 @@ namespace glm
/// @param near
/// @tparam T Value type used to build the matrix. Currently supported: half (not recommanded), float or double.
/// @see gtc_matrix_transform
template <typename T>
template <typename T>
detail::tmat4x4<T> infinitePerspective(
T fovy, T aspect, T near);
@@ -220,7 +220,7 @@ namespace glm
/// @param near
/// @tparam T Value type used to build the matrix. Currently supported: half (not recommanded), float or double.
/// @see gtc_matrix_transform
template <typename T>
template <typename T>
detail::tmat4x4<T> tweakedInfinitePerspective(
T fovy, T aspect, T near);

8
glm/gtc/matrix_transform.inl
View File

@@ -236,7 +236,13 @@ namespace glm
valType const & zFar
)
{
valType range = tan(radians(fovy / valType(2))) * zNear;
#ifdef GLM_FORCE_RADIANS
valType const rad = fovy;
#else
valType const rad = glm::radians(fovy);
#endif
valType range = tan(radians(rad / valType(2))) * zNear;
valType left = -range * aspect;
valType right = range * aspect;
valType bottom = -range;

45
glm/gtc/quaternion.hpp
View File

@@ -22,12 +22,13 @@
///
/// @ref gtc_quaternion
/// @file glm/gtc/quaternion.hpp
/// @date 2009-05-21 / 2011-06-05
/// @date 2009-05-21 / 2012-12-20
/// @author Christophe Riccio
///
/// @see core (dependence)
/// @see gtc_half_float (dependence)
///
/// @see gtc_constants (dependence)
///
/// @defgroup gtc_quaternion GLM_GTC_quaternion
/// @ingroup gtc
///
@@ -42,6 +43,7 @@
// Dependency:
#include "../glm.hpp"
#include "../gtc/half_float.hpp"
#include "../gtc/constants.hpp"
#if(defined(GLM_MESSAGES) && !defined(glm_ext))
# pragma message("GLM: GLM_GTC_quaternion extension included")
@@ -169,15 +171,50 @@ namespace detail
detail::tquat<T> const & q1,
detail::tquat<T> const & q2);
/// Returns a SLERP interpolated quaternion of x and y according a.
/// Spherical linear interpolation of two quaternions.
/// The interpolation is oriented and the rotation is performed at constant speed.
/// For short path spherical linear interpolation, use the slerp function.
///
/// @param x A quaternion
/// @param y A quaternion
/// @param a Interpolation factor. The interpolation is defined beyond the range [0, 1].
/// @tparam T Value type used to build the quaternion. Supported: half, float or double.
/// @see gtc_quaternion
/// @see - slerp(detail::tquat<T> const & x, detail::tquat<T> const & y, T const & a)
template <typename T>
detail::tquat<T> mix(
detail::tquat<T> const & x,
detail::tquat<T> const & y,
T const & a);
/// Linear interpolation of two quaternions.
/// The interpolation is oriented.
///
/// @param x A quaternion
/// @param y A quaternion
/// @param a Interpolation factor. The interpolation is defined in the range [0, 1].
/// @tparam T Value type used to build the quaternion. Supported: half, float or double.
/// @see gtc_quaternion
template <typename T>
detail::tquat<T> lerp(
detail::tquat<T> const & x,
detail::tquat<T> const & y,
T const & a);
/// Spherical linear interpolation of two quaternions.
/// The interpolation always take the short path and the rotation is performed at constant speed.
///
/// @param x A quaternion
/// @param y A quaternion
/// @param a Interpolation factor. The interpolation is defined beyond the range [0, 1].
/// @tparam T Value type used to build the quaternion. Supported: half, float or double.
/// @see gtc_quaternion
template <typename T>
detail::tquat<T> slerp(
detail::tquat<T> const & x,
detail::tquat<T> const & y,
T const & a);
/// Returns the q conjugate.
///
/// @see gtc_quaternion

80
glm/gtc/quaternion.inl
View File

@@ -443,6 +443,7 @@ namespace detail
return normalize(beta * x + alpha * y);
}
*/
template <typename T>
GLM_FUNC_QUALIFIER detail::tquat<T> mix
(
@@ -451,8 +452,77 @@ namespace detail
T const & a
)
{
T angle = acos(dot(x, y));
return (glm::sin((T(1) - a) * angle) * x + glm::sin(a * angle) * y) / glm::sin(angle);
T cosTheta = dot(x, y);
// Perform a linear interpolation when cosTheta is close to 1 to avoid side effect of sin(angle) becoming a zero denominator
if(cosTheta > T(1) - epsilon<T>())
{
// Linear interpolation
return detail::tquat<T>(
mix(x.w, y.w, a),
mix(x.x, y.x, a),
mix(x.y, y.y, a),
mix(x.z, y.z, a));
}
else
{
// Essential Mathematics, page 467
T angle = acos(cosTheta);
return (sin((T(1) - a) * angle) * x + sin(a * angle) * y) / sin(angle);
}
}
template <typename T>
GLM_FUNC_QUALIFIER detail::tquat<T> lerp
(
detail::tquat<T> const & x,
detail::tquat<T> const & y,
T const & a
)
{
// Lerp is only defined in [0, 1]
assert(a >= T(0));
assert(a <= T(1));
return x * (T(1) - a) + (y * a);
}
template <typename T>
GLM_FUNC_QUALIFIER detail::tquat<T> slerp
(
detail::tquat<T> const & x,
detail::tquat<T> const & y,
T const & a
)
{
detail::tquat<T> z = y;
T cosTheta = dot(x, y);
// If cosTheta < 0, the interpolation will take the long way around the sphere.
// To fix this, one quat must be negated.
if (cosTheta < T(0))
{
z = -y;
cosTheta = -cosTheta;
}
// Perform a linear interpolation when cosTheta is close to 1 to avoid side effect of sin(angle) becoming a zero denominator
if(cosTheta > T(1) - epsilon<T>())
{
// Linear interpolation
return detail::tquat<T>(
mix(x.w, y.w, a),
mix(x.x, y.x, a),
mix(x.y, y.y, a),
mix(x.z, y.z, a));
}
else
{
// Essential Mathematics, page 467
T angle = acos(cosTheta);
return (sin((T(1) - a) * angle) * x + sin(a * angle) * z) / sin(angle);
}
}
template <typename T>
@@ -592,7 +662,7 @@ namespace detail
typename detail::tquat<T>::value_type fourYSquaredMinus1 = m[1][1] - m[0][0] - m[2][2];
typename detail::tquat<T>::value_type fourZSquaredMinus1 = m[2][2] - m[0][0] - m[1][1];
typename detail::tquat<T>::value_type fourWSquaredMinus1 = m[0][0] + m[1][1] + m[2][2];
int biggestIndex = 0;
typename detail::tquat<T>::value_type fourBiggestSquaredMinus1 = fourWSquaredMinus1;
if(fourXSquaredMinus1 > fourBiggestSquaredMinus1)
@@ -641,6 +711,10 @@ namespace detail
Result.y = (m[1][2] + m[2][1]) * mult;
Result.z = biggestVal;
break;
default: // Silence a -Wswitch-default warning in GCC. Should never actually get here. Assert is just for sanity.
assert(false);
break;
}
return Result;
}

10
glm/gtc/ulp.inl
View File

@@ -36,8 +36,10 @@
#include <cmath>
#include <cfloat>
#pragma warning(push)
#pragma warning(disable : 4127)
#if(GLM_COMPILER & GLM_COMPILER_VC)
# pragma warning(push)
# pragma warning(disable : 4127)
#endif
typedef union
{
@@ -186,7 +188,9 @@ namespace detail
}//namespace detail
}//namespace glm
#pragma warning(pop)
#if(GLM_COMPILER & GLM_COMPILER_VC)
# pragma warning(pop)
#endif
#if((GLM_COMPILER & GLM_COMPILER_VC) || ((GLM_COMPILER & GLM_COMPILER_INTEL) && (GLM_PLATFORM & GLM_PLATFORM_WINDOWS)))
# define GLM_NEXT_AFTER_FLT(x, toward) glm::detail::nextafterf((x), (toward))

2
glm/gtx/component_wise.inl
View File

@@ -12,7 +12,7 @@ namespace glm
template <typename genType>
GLM_FUNC_QUALIFIER typename genType::value_type compAdd(genType const & v)
{
typename genType::size_type result = typename genType::value_type(0);
typename genType::value_type result = typename genType::value_type(0);
for(typename genType::size_type i = 0; i < v.length(); ++i)
result += v[i];
return result;

2
glm/gtx/norm.hpp
View File

@@ -31,7 +31,7 @@
/// @defgroup gtx_norm GLM_GTX_norm
/// @ingroup gtx
///
/// @brief Various way to compute vector norms.
/// @brief Various ways to compute vector norms.
///
/// <glm/gtx/norm.hpp> need to be included to use these functionalities.
///////////////////////////////////////////////////////////////////////////////////

2
glm/gtx/simd_mat4.hpp
View File

@@ -91,6 +91,8 @@ namespace detail
fvec4SIMD const & v3);
explicit fmat4x4SIMD(
tmat4x4<float> const & m);
explicit fmat4x4SIMD(
__m128 const in[4]);
// Conversions
//template <typename U>

292
glm/gtx/simd_mat4.inl
View File

@@ -26,7 +26,14 @@ GLM_FUNC_QUALIFIER fmat4x4SIMD::size_type fmat4x4SIMD::row_size()
}
GLM_FUNC_QUALIFIER fmat4x4SIMD::fmat4x4SIMD()
{}
{
#ifndef GLM_SIMD_ENABLE_DEFAULT_INIT
this->Data[0] = fvec4SIMD(1.0f, 0, 0, 0);
this->Data[1] = fvec4SIMD(0, 1.0f, 0, 0);
this->Data[2] = fvec4SIMD(0, 0, 1.0f, 0);
this->Data[3] = fvec4SIMD(0, 0, 0, 1.0f);
#endif
}
GLM_FUNC_QUALIFIER fmat4x4SIMD::fmat4x4SIMD(float const & s)
{
@@ -75,6 +82,17 @@ GLM_FUNC_QUALIFIER fmat4x4SIMD::fmat4x4SIMD
this->Data[3] = fvec4SIMD(m[3]);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD::fmat4x4SIMD
(
__m128 const in[4]
)
{
this->Data[0] = in[0];
this->Data[1] = in[1];
this->Data[2] = in[2];
this->Data[3] = in[3];
}
//////////////////////////////////////
// Accesses
@@ -84,7 +102,7 @@ GLM_FUNC_QUALIFIER fvec4SIMD & fmat4x4SIMD::operator[]
)
{
assert(
i >= fmat4x4SIMD::size_type(0) &&
//i >= fmat4x4SIMD::size_type(0) &&
i < fmat4x4SIMD::col_size());
return this->Data[i];
@@ -96,7 +114,7 @@ GLM_FUNC_QUALIFIER fvec4SIMD const & fmat4x4SIMD::operator[]
) const
{
assert(
i >= fmat4x4SIMD::size_type(0) &&
//i >= fmat4x4SIMD::size_type(0) &&
i < fmat4x4SIMD::col_size());
return this->Data[i];
@@ -232,6 +250,274 @@ GLM_FUNC_QUALIFIER fmat4x4SIMD & fmat4x4SIMD::operator-- ()
return *this;
}
//////////////////////////////////////////////////////////////
// Binary operators
GLM_FUNC_QUALIFIER fmat4x4SIMD operator+
(
const fmat4x4SIMD &m,
float const & s
)
{
return detail::fmat4x4SIMD
(
m[0] + s,
m[1] + s,
m[2] + s,
m[3] + s
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator+
(
float const & s,
const fmat4x4SIMD &m
)
{
return detail::fmat4x4SIMD
(
m[0] + s,
m[1] + s,
m[2] + s,
m[3] + s
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator+
(
const fmat4x4SIMD &m1,
const fmat4x4SIMD &m2
)
{
return detail::fmat4x4SIMD
(
m1[0] + m2[0],
m1[1] + m2[1],
m1[2] + m2[2],
m1[3] + m2[3]
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator-
(
const fmat4x4SIMD &m,
float const & s
)
{
return detail::fmat4x4SIMD
(
m[0] - s,
m[1] - s,
m[2] - s,
m[3] - s
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator-
(
float const & s,
const fmat4x4SIMD &m
)
{
return detail::fmat4x4SIMD
(
s - m[0],
s - m[1],
s - m[2],
s - m[3]
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator-
(
const fmat4x4SIMD &m1,
const fmat4x4SIMD &m2
)
{
return detail::fmat4x4SIMD
(
m1[0] - m2[0],
m1[1] - m2[1],
m1[2] - m2[2],
m1[3] - m2[3]
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator*
(
const fmat4x4SIMD &m,
float const & s
)
{
return detail::fmat4x4SIMD
(
m[0] * s,
m[1] * s,
m[2] * s,
m[3] * s
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator*
(
float const & s,
const fmat4x4SIMD &m
)
{
return detail::fmat4x4SIMD
(
m[0] * s,
m[1] * s,
m[2] * s,
m[3] * s
);
}
GLM_FUNC_QUALIFIER fvec4SIMD operator*
(
const fmat4x4SIMD &m,
fvec4SIMD const & v
)
{
return sse_mul_ps(&m.Data[0].Data, v.Data);
}
GLM_FUNC_QUALIFIER fvec4SIMD operator*
(
fvec4SIMD const & v,
const fmat4x4SIMD &m
)
{
return sse_mul_ps(v.Data, &m.Data[0].Data);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator*
(
const fmat4x4SIMD &m1,
const fmat4x4SIMD &m2
)
{
fmat4x4SIMD result;
sse_mul_ps(&m1.Data[0].Data, &m2.Data[0].Data, &result.Data[0].Data);
return result;
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator/
(
const fmat4x4SIMD &m,
float const & s
)
{
return detail::fmat4x4SIMD
(
m[0] / s,
m[1] / s,
m[2] / s,
m[3] / s
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator/
(
float const & s,
const fmat4x4SIMD &m
)
{
return detail::fmat4x4SIMD
(
s / m[0],
s / m[1],
s / m[2],
s / m[3]
);
}
GLM_FUNC_QUALIFIER fvec4SIMD operator/
(
const fmat4x4SIMD &m,
fvec4SIMD const & v
)
{
return inverse(m) * v;
}
GLM_FUNC_QUALIFIER fvec4SIMD operator/
(
fvec4SIMD const & v,
const fmat4x4SIMD &m
)
{
return v * inverse(m);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD operator/
(
const fmat4x4SIMD &m1,
const fmat4x4SIMD &m2
)
{
__m128 result[4];
__m128 inv[4];
sse_inverse_ps(&m2.Data[0].Data, inv);
sse_mul_ps(&m1.Data[0].Data, inv, result);
return fmat4x4SIMD(result);
}
//////////////////////////////////////////////////////////////
// Unary constant operators
GLM_FUNC_QUALIFIER fmat4x4SIMD const operator-
(
fmat4x4SIMD const & m
)
{
return detail::fmat4x4SIMD
(
-m[0],
-m[1],
-m[2],
-m[3]
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD const operator--
(
fmat4x4SIMD const & m,
int
)
{
return detail::fmat4x4SIMD
(
m[0] - 1.0f,
m[1] - 1.0f,
m[2] - 1.0f,
m[3] - 1.0f
);
}
GLM_FUNC_QUALIFIER fmat4x4SIMD const operator++
(
fmat4x4SIMD const & m,
int
)
{
return detail::fmat4x4SIMD
(
m[0] + 1.0f,
m[1] + 1.0f,
m[2] + 1.0f,
m[3] + 1.0f
);
}
}//namespace detail
GLM_FUNC_QUALIFIER detail::tmat4x4<float> mat4_cast

28
glm/gtx/simd_vec4.hpp
View File

@@ -54,6 +54,14 @@
# pragma message("GLM: GLM_GTX_simd_vec4 extension included")
#endif
// Warning silencer for nameless struct/union.
#if (GLM_COMPILER & GLM_COMPILER_VC)
# pragma warning(push)
# pragma warning(disable:4201) // warning C4201: nonstandard extension used : nameless struct/union
#endif
namespace glm{
namespace detail
{
@@ -69,7 +77,15 @@ namespace detail
typedef fvec4SIMD type;
typedef tvec4<bool> bool_type;
__m128 Data;
#ifdef GLM_SIMD_ENABLE_XYZW_UNION
union
{
__m128 Data;
struct {float x, y, z, w;};
};
#else
__m128 Data;
#endif
//////////////////////////////////////
// Implicit basic constructors
@@ -289,8 +305,8 @@ namespace detail
//! you would want a threshold function with a smooth
//! transition. This is equivalent to:
//! genType t;
//! t = clamp ((x <EFBFBD> edge0) / (edge1 <EFBFBD> edge0), 0, 1);
//! return t * t * (3 <EFBFBD> 2 * t);
//! t = clamp ((x - edge0) / (edge1 - edge0), 0, 1);
//! return t * t * (3 - 2 * t);
//! Results are undefined if edge0 >= edge1.
//! (From GLM_GTX_simd_vec4 extension, common function)
detail::fvec4SIMD smoothstep(
@@ -490,6 +506,12 @@ namespace detail
#include "simd_vec4.inl"
#if (GLM_COMPILER & GLM_COMPILER_VC)
# pragma warning(pop)
#endif
#endif//(GLM_ARCH != GLM_ARCH_PURE)
#endif//GLM_GTX_simd_vec4

5
glm/gtx/simd_vec4.inl
View File

@@ -20,6 +20,9 @@ struct mask
// Implicit basic constructors
GLM_FUNC_QUALIFIER fvec4SIMD::fvec4SIMD()
#ifdef GLM_SIMD_ENABLE_DEFAULT_INIT
: Data(_mm_set_ps(0.0f, 0.0f, 0.0f, 0.0f))
#endif
{}
GLM_FUNC_QUALIFIER fvec4SIMD::fvec4SIMD(__m128 const & Data) :
@@ -449,7 +452,7 @@ GLM_FUNC_QUALIFIER detail::fvec4SIMD mix
{
__m128 Sub0 = _mm_sub_ps(y.Data, x.Data);
__m128 Mul0 = _mm_mul_ps(a.Data, Sub0);
return _mm_mul_ps(x.Data, Mul0);
return _mm_add_ps(x.Data, Mul0);
}
GLM_FUNC_QUALIFIER detail::fvec4SIMD step